Maternal vaccination with the 2009 pH1N1 influenza vaccine and risk of spontaneous abortion
EBM Focus - Volume 12, Issue 40
Reference: Vaccine 2017 Sep 25;35(40):5314
- Influenza vaccination is recommended for pregnant women. Numerous studies have demonstrated maternal and fetal safety, but evidence during early pregnancy is relatively limited.
- A recent case-control study assessed possible associations between maternal vaccination with an inactive influenza vaccine containing the 2009 pandemic H1N1 antigens and subsequent spontaneous abortion (SAB) in 485 women who had a SAB and 485 matched controls.
- A statistical association was found between having had a SAB and having had a pH1N1 influenza vaccine within the previous 28 days. However, numerous factors limit interpretation and clinical relevance and this study should not be used to recommend against influenza vaccination while pregnant.
The United States Advisory Committee on Immunization Practices recommends influenza vaccination for pregnant women due to an increased risk of complications and severe illness (MMWR 2017 Aug). Many studies have demonstrated maternal and fetal safety with flu vaccination (Vaccine 2015, Clin Infect Dis 2015, BJOG 2015), but evidence during early pregnancy is limited (MMWR 2017 Aug). A recent case-control study used electronic health records within a system of healthcare providers in the United States to assess possible associations between maternal vaccination with an inactive influenza vaccine containing the 2009 pandemic H1N1 antigens (pH1N1 vaccine) and subsequent SAB (fetal death at gestational age 5 to < 20 weeks). The study focused on events occurring during influenza seasons in 2010-2011 and 2011-2012. In the study, 485 women aged 18-44 years who had a clinically or imaging-confirmed SAB during the influenza seasons were compared with matched controls who were women who had a live birth or stillbirth (fetal death at gestational age ≥ 20 weeks). The controls were matched by date of last menstrual period, maternal age group (< 30 years vs. ≥ 30 years old), and healthcare organization. Women who were not continually enrolled with the healthcare organization for ≥ 12 months were excluded. The reference date for each matched pair was the estimated date of the SAB, and the likelihood of having had a previous pH1N1 vaccination was compared in women with SAB to the matched controls. Three time periods of vaccination before the reference date were investigated: pH1N1 vaccination within 28 days prior to the reference date (recorded in 62 women [6% of the cohort]), 29-56 days prior (in 33 women [3%]), and more than 56 days prior (in 155 women [16%]).
Compared to women who did not have a SAB, women who had a SAB were more likely to have had the pH1N1 vaccine within 28 days prior to the reference date (adjusted odds ratio 2, 95% CI 1.1-3.6). The results were consistent in post-hoc analyses for women assessed during 2010-2011, but not 2011-2012, and for women who had the vaccine during the previous year, but not among women who did not. The likelihood of having had the pH1N1 vaccine 29-56 days or > 56 days prior to the reference date was not significantly greater among women who had a SAB, with consistent results for each season analyzed separately and regardless of vaccinations during the previous year.
This case-control study found a statistical association between having had a SAB and having had a pH1N1 vaccine within the prior 28 days, but interpretation of this finding is limited by several factors. First, a case-control study has greater bias than a cohort study, which comprises the most common type of study included in several recent systematic reviews finding no association between maternal influenza vaccines (of different compositions) and poor fetal outcomes (Vaccine 2015, Clin Infect Dis 2015, BJOG 2015). Second, the association between SAB and pH1N1 vaccination within the prior 28 days was not consistent among different subgroups of women in post-hoc analyses. Third, the total number of women who had the vaccine within 28 days prior to the reference date was small (only 62 women, 21 of whom also had the vaccine during the previous year). The latter two factors raise the question of whether the observed statistical association represents a true connection or a statistical artifact. Fourth, a possible confounder that was not explicitly addressed is that women who get a flu vaccine may be more likely to go to their healthcare provider when they have a SAB. Analytic techniques such as the use of propensity scores to guide control selection may have addressed this bias. Also, vaccinations not recorded in the electronic health records, such as those provided by an employer or at a clinic, were not included.
Finally, previous studies found no association between different flu vaccines and poor fetal outcomes, while this study found an association between a specific vaccine and SAB for select women. The relevance of that association to subsequent vaccines with different compositions is unclear (in fact, different strains comprise this year’s recommended vaccine, WHO 2017 Mar, MMWR 2017 Aug). It is also unclear how the potential association found in this case-control study (which cannot estimate the absolute risk of having a SAB after a vaccination) compares with associations between maternal influenza virus infection and poor maternal, fetal, and birth outcomes, including fetal death (BJOG 2017, Hum Reprod 2014, Obstet Gynecol Clin North Am 2015, Am J Obstet Gynecol 2012). The safety of any pharmaceutical product is important and this study provides hypothesis-generating results to motivate more rigorous studies such as high-quality cohort studies. The results from this study should be interpreted with caution and should not be used to recommend against influenza vaccination while pregnant.